wbmw, chipguy, I'm not really confusing architecture and design. A design is ultimately limited by the architecture upon which it is based, and if the architecture of x86 was as bad as people say then the great designs would not have been possible.
Really, I am amused by another post I saw here today talking about (to paraphase) x86 violating any good design taught in the universities. There is a lot of accademic envy in that assertion!
(Kind of reminds me of the OSI network driver reference model, which was an accademic exercise in defining seven layers of network connectivity - each one that must speak to a peer in the pure model. 3Com actually made an OSI protocol stack once, ran horribly. Anyone sensible could have predicted all the extra data handling and signalling required!)
University professors like nice, symmetrical architectures which can then be used for their nice Unix systems. Their biases for internal beauty fails when the rubber meets the road - all their objections to the x86 architecture turned out to be simply short-sighted problems based on the state of semiconductor integration back in the 1980s. With the much higher degree of integration now, solutions were found for these problems, and the x86 runs circles around all the elegant RISC processors they preferred.
As you pointed out, much of the legacy code is now purged - you do not have to support real mode applications, or even 286 protected mode. In 64-bit mode, I don't think that AMD64 even supports that stuff! In 32-bit mode it does, but I recall that 64-bit mode is only backwards compatible to x86-32, which is not a bad architecture at all.
Note to chipguy: You really don't think that x86 is responsible for the displacement of all those other architectures? What a warped view of history! As you said, MIPS is still going strong in the embedded control market - but you forget that this is the architecture that WindowsNT originally was based on (after Microsoft gave up on the Intel NT architecture, the i860), and that mips was intended as a desktop RISC processor. VAX were buried by RISCs (primarily SPARC), and SPARC lost the departmental server & workstation business to x86 - they tried to go upscale with it, but IBM is beating the pants off this repositioning. Do you forget how jealous McNealy has always been of Gates since Sun could never penetrate the desktop, in spite of their efforts? You have a partial point with HP, although they saw the handwriting of an Intel-dominated world and that had quite a bit to do with their decision.
Market Data 
Markets 
AI
